I have data with columns like this:
ENTITY Avg_Score_1 N_obs_1 Avg_Score_2 N_obs_2
With sample values:
001 | 0.997 | 900 | 1.13 | 905
002 | 0.890 | 250 | 0.96 | 251
For about 1000 unique ENTITY values, and I’m wondering how to calculate the ICC between avg_score_1 and avg_score_2 for the whole dataset.
Specifically I’m confused about where exactly each mean square value should come from. If necessary I could also get the standard deviation for each score on each row as well. Are the N_obs columns necessary here? It seems like they should be but all the other resources I’ve looked at online seems to assume each score is based on a single observation
Thanks